正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

DynamicVLA A compact 0.4B Vision-Language-Action model that finally lets robots manipulate moving objects in real-time, closing the perception-execution gap with Continuous Inference and Latent-aware Action Streaming.

DailyPapers

22,198 subscribers

16,396 次观看 • 9 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

5 条评论

DailyPapers 的头像

DailyPapers9 个月前

See it in action catching rolling tennis balls and bottles. New DOM benchmark (200K+ episodes) evaluates perception, interaction, and generalization. Project: Paper: Dataset:

Tobe Duru 的头像

Tobe Duru9 个月前

the future of robotic manipulation is here

Dhruv Diddi 的头像

Dhruv Diddi9 个月前

Cute 😅

House of Phill 的头像

House of Phill9 个月前

Catching rolling objects in real time is a big step beyond static tabletop VLAs. Curious how DynamicVLA scales across embodiments and whether similar gains hold outside the DOM setup.

Bixente 的头像

Bixente9 个月前

Where model, dataset I don't care

相关视频

Introducing TraceVLA: a fully open-source Vision-Language-Action model reimagining spatial-temporal awareness: ✨ 3.5x gains on real robots, SOTA in simulation 💡 Fine-tunes on just 150K trajectories ⚡ Compact 4B model = 7B performance

Introducing TraceVLA: a fully open-source Vision-Language-Action model reimagining spatial-temporal awareness: ✨ 3.5x gains on real robots, SOTA in simulation 💡 Fine-tunes on just 150K trajectories ⚡ Compact 4B model = 7B performance

Yongyuan Liang

39,603 次观看 • 1 年前

Chinese robotics company Astribot released their latest World-Action Model (WAM), Lumo-2. Technical breakdown: - based on a frozen 🥶 Qwen-3.5 4B VLM - trained in 3 progressive stages: 1. Action is aligned with latent world dynamics (an abstract representation of action). Real-world actions are anchored to physical constraints, while the latent space is guided to focus on motion-relevant changes. This bidirectional relationship makes the model physically grounded -> critical for a world model. 2. Action is aligned with vision and language. Reusing the vision backbone and action encoder from the frozen VLM, the authors add a custom vocabulary (for new actions), a semantic module, an action decoder, and an action projector. This aligns the (new) action representations with the (existing) vision-language semantic space. Most importantly: it builds a direct mapping from natural-language instructions to motor execution. 3. End-to-end training on language, video, and robot data. Only the new modules (everything outside the frozen backbone) are trained end-to-end across temporal reasoning, physical understanding, long-horizon, and dexterous manipulation. At the end of the day, Lumo-2 is not the best on benchmarks, but that's not the point. What's genuinely new: - a way to combine latent world modeling and action generation through progressive alignment - a physically-grounded latent dynamics space - it lifts performance on unseen objects using un-annotated human egocentric video + Vision Pro captures, no special transfer algorithm needed Why it matters: - the whole model is thin trainable adapters (semantic module, action decoder/projector) on a frozen 4B backbone (cheap) - that scale is suited for real-time embedded inference (~2.71× decode speedup, no accuracy loss) - its real moat is long-horizon execution, where the added temporal memory pays off far more than on any other task As a result, this robot can now make your latte (5x sped up video):

Chinese robotics company Astribot released their latest World-Action Model (WAM), Lumo-2. Technical breakdown: - based on a frozen 🥶 Qwen-3.5 4B VLM - trained in 3 progressive stages: 1. Action is aligned with latent world dynamics (an abstract representation of action). Real-world actions are anchored to physical constraints, while the latent space is guided to focus on motion-relevant changes. This bidirectional relationship makes the model physically grounded -> critical for a world model. 2. Action is aligned with vision and language. Reusing the vision backbone and action encoder from the frozen VLM, the authors add a custom vocabulary (for new actions), a semantic module, an action decoder, and an action projector. This aligns the (new) action representations with the (existing) vision-language semantic space. Most importantly: it builds a direct mapping from natural-language instructions to motor execution. 3. End-to-end training on language, video, and robot data. Only the new modules (everything outside the frozen backbone) are trained end-to-end across temporal reasoning, physical understanding, long-horizon, and dexterous manipulation. At the end of the day, Lumo-2 is not the best on benchmarks, but that's not the point. What's genuinely new: - a way to combine latent world modeling and action generation through progressive alignment - a physically-grounded latent dynamics space - it lifts performance on unseen objects using un-annotated human egocentric video + Vision Pro captures, no special transfer algorithm needed Why it matters: - the whole model is thin trainable adapters (semantic module, action decoder/projector) on a frozen 4B backbone (cheap) - that scale is suited for real-time embedded inference (~2.71× decode speedup, no accuracy loss) - its real moat is long-horizon execution, where the added temporal memory pays off far more than on any other task As a result, this robot can now make your latte (5x sped up video):

Léo

32,513 次观看 • 3 个月前

🚀🚀🚀 Ever wondered what it takes for robots to handle real-world household tasks? long-horizon execution, deformable object dexterity, and unseen object generalization — meet GR-3, ByteDance Seed’s new Vision-Language-Action (VLA) model! GR-3 is a generalizable Vision-Language-Action (VLA) model with strong capabilities in complex long-horizon tasks. It understands unseen abstract concepts, manipulates deformable objects robustly, and adapts to novel settings with minimal human data. ✨ Generalization: Generalizes well to unseen objects, environments, and even instructions with abstract concepts. ✨ Long-Horizon Manipulation: Completes long-horizon tasks with strong instruction-following capabilities. ✨ Deformable Object Manipulation: Manipulate deformable objects robustly. Project Page: Arxiv: #ByteDance #ByteDanceSeed #GR3 #VLA #Robotics #FoundationModels

🚀🚀🚀 Ever wondered what it takes for robots to handle real-world household tasks? long-horizon execution, deformable object dexterity, and unseen object generalization — meet GR-3, ByteDance Seed’s new Vision-Language-Action (VLA) model! GR-3 is a generalizable Vision-Language-Action (VLA) model with strong capabilities in complex long-horizon tasks. It understands unseen abstract concepts, manipulates deformable objects robustly, and adapts to novel settings with minimal human data. ✨ Generalization: Generalizes well to unseen objects, environments, and even instructions with abstract concepts. ✨ Long-Horizon Manipulation: Completes long-horizon tasks with strong instruction-following capabilities. ✨ Deformable Object Manipulation: Manipulate deformable objects robustly. Project Page: Arxiv: #ByteDance #ByteDanceSeed #GR3 #VLA #Robotics #FoundationModels

Xiao Ma

46,482 次观看 • 1 年前

Tired of your vision-language-action (VLA) model failing catastrophically in the presence of distractions? Check out BYOVLA: Bring Your Own VLA: a run-time intervention scheme that markedly improves performance with distractor objects and backgrounds.

Tired of your vision-language-action (VLA) model failing catastrophically in the presence of distractions? Check out BYOVLA: Bring Your Own VLA: a run-time intervention scheme that markedly improves performance with distractor objects and backgrounds.

Anirudha Majumdar

17,831 次观看 • 2 年前

Check out this Circulus Pion robot at Computex 2026. It’s running on a Panther Lake processor, using P-cores for real-time control, E-cores to collect sensor data, the GPU for vision-language-action models, and the NPU for continuous perception.

Check out this Circulus Pion robot at Computex 2026. It’s running on a Panther Lake processor, using P-cores for real-time control, E-cores to collect sensor data, the GPU for vision-language-action models, and the NPU for continuous perception.

Intel Technology

10,752 次观看 • 5 个月前

New paper introduces NaVILA, a vision-language-action (VLA) model that integrates high-level visual-language understanding and low-level locomotion control. It enables humanoid or quadruped robots to navigate unseen environments with natural language instructions.

New paper introduces NaVILA, a vision-language-action (VLA) model that integrates high-level visual-language understanding and low-level locomotion control. It enables humanoid or quadruped robots to navigate unseen environments with natural language instructions.

The Humanoid Hub

20,484 次观看 • 1 年前

🚀 MiniCPM enters the physical world — enabling robots to understand, remember, and act. We open-source MiniCPM-Robot, our first embodied AI model series, including: 🤖 MiniCPM-RobotManip — a 1.5B general-purpose Vision-Language-Action (VLA) model for robotic manipulation. 🐕 MiniCPM-RobotTrack — a compact model for real-world target tracking. ⚡ PhyAI — a high-performance inference framework built for embodied models. Together, they bring efficient, practical, and open embodied intelligence closer to real-world robots. ⭐ GitHub: 🤗 MiniCPM-RobotManip: 🤗 MiniCPM-RobotTrack:

🚀 MiniCPM enters the physical world — enabling robots to understand, remember, and act. We open-source MiniCPM-Robot, our first embodied AI model series, including: 🤖 MiniCPM-RobotManip — a 1.5B general-purpose Vision-Language-Action (VLA) model for robotic manipulation. 🐕 MiniCPM-RobotTrack — a compact model for real-world target tracking. ⚡ PhyAI — a high-performance inference framework built for embodied models. Together, they bring efficient, practical, and open embodied intelligence closer to real-world robots. ⭐ GitHub: 🤗 MiniCPM-RobotManip: 🤗 MiniCPM-RobotTrack:

OpenBMB

332,156 次观看 • 3 个月前

Physicalomous is a robot powered by a Vision-Language-Action model that can clean homes autonomously, adapting to unfamiliar environments in real time without needing prior training in each home.

Physicalomous is a robot powered by a Vision-Language-Action model that can clean homes autonomously, adapting to unfamiliar environments in real time without needing prior training in each home.

Science girl

49,725 次观看 • 22 天前

Robots movements are getting better at a scary pace. Huge real-world datasets and VLA models now turn vision-language into smooth continuous control, diffusion policies generate coherent action sequences, and compliant sensing-actuation reduces jitter.

Robots movements are getting better at a scary pace. Huge real-world datasets and VLA models now turn vision-language into smooth continuous control, diffusion policies generate coherent action sequences, and compliant sensing-actuation reduces jitter.

Rohan Paul

124,561 次观看 • 11 个月前

Large language models reason through text. Vision‑language‑action models reason through the real world. By fusing perception, context, and action from live video, VLAs deliver the awareness physical AI needs for next‑gen robotics and edge systems.

Large language models reason through text. Vision‑language‑action models reason through the real world. By fusing perception, context, and action from live video, VLAs deliver the awareness physical AI needs for next‑gen robotics and edge systems.

Intel

15,931 次观看 • 9 个月前

Chain-of-thought reasoning is a powerful tool to enable language models to work through complex problems. Can we use this with robots? With embodied chain-of-thought, vision-language-action (VLA) models can think through perception and planning! A 🧵👇

Chain-of-thought reasoning is a powerful tool to enable language models to work through complex problems. Can we use this with robots? With embodied chain-of-thought, vision-language-action (VLA) models can think through perception and planning! A 🧵👇

Sergey Levine

30,485 次观看 • 2 年前

3D-VLA A 3D Vision-Language-Action Generative World Model Recent vision-language-action (VLA) models rely on 2D inputs, lacking integration with the broader realm of the 3D physical world. Furthermore, they perform action prediction by learning a direct mapping from

3D-VLA A 3D Vision-Language-Action Generative World Model Recent vision-language-action (VLA) models rely on 2D inputs, lacking integration with the broader realm of the 3D physical world. Furthermore, they perform action prediction by learning a direct mapping from

AK

85,570 次观看 • 2 年前

1/N Most Vision-Language-Action models need tons of data for finetuning, and still fail for new objects and instructions. Introducing OTTER, a lightweight, easy-to-train model that uses text-aware visual features to nail unseen tasks out of the box! Here's how it works 👇

1/N Most Vision-Language-Action models need tons of data for finetuning, and still fail for new objects and instructions. Introducing OTTER, a lightweight, easy-to-train model that uses text-aware visual features to nail unseen tasks out of the box! Here's how it works 👇

Fangchen Liu

69,453 次观看 • 1 年前

A new paradigm for robot learning that unifies latent world modeling and action generation. Astribot has introduced Lumo-2, a latent world-action model for robot control. - Validated on a real robot through 20+ intricate chores, extending beyond simple benchmark testing. - A 3-stage recipe that progressively aligns actions with world dynamics, then vision, then language - Action quality depends on the geometry of the latent space, not just reconstruction accuracy - BAR (block-wise decoding) speeds up inference by 2.71x compared to standard autoregressive decoding - Shows cross-embodiment transfer without specialized transfer learning Results: - Beats π0.5 and Fast-WAM in temporal reasoning, physical understanding, long-horizon, and dexterous manipulation - Lumo-2's VLM consistently outperforms Qwen-3.5 4B on embodied benchmarks

A new paradigm for robot learning that unifies latent world modeling and action generation. Astribot has introduced Lumo-2, a latent world-action model for robot control. - Validated on a real robot through 20+ intricate chores, extending beyond simple benchmark testing. - A 3-stage recipe that progressively aligns actions with world dynamics, then vision, then language - Action quality depends on the geometry of the latent space, not just reconstruction accuracy - BAR (block-wise decoding) speeds up inference by 2.71x compared to standard autoregressive decoding - Shows cross-embodiment transfer without specialized transfer learning Results: - Beats π0.5 and Fast-WAM in temporal reasoning, physical understanding, long-horizon, and dexterous manipulation - Lumo-2's VLM consistently outperforms Qwen-3.5 4B on embodied benchmarks

The Humanoid Hub

20,623 次观看 • 3 个月前

World models have a causality problem. Realistic videos are not enough. A world model should predict the future caused by an action, not just a plausible future. We find that many latent-action world models generate convincing videos while barely responding to the supplied action. The root cause lies in Latent Action Models (LAMs): reconstruction objectives entangle action-induced dynamics with action-irrelevant visual changes, producing visually confounded latent actions. Our solution is CD-LAM (Causally Debiased Latent Action Model). It cleanses latent action representations before pre-training begins, requiring zero changes to backbone architectures, latent dimensions, or action interfaces. Key Benchmark Results: - Action Controllability: Cuts action-following error by >30% while simultaneously enhancing visual fidelity. - Extreme Sample Efficiency: Achieves baseline performance in just 3,000 post-training steps instead of 50,000 - a >10x speedup. By removing visual confounding, CD-LAM bridges the gap between passive video prediction and actionable causal intelligence for robotics. The next bottleneck for world models isn’t realism. It’s causality. Paper, project page and additional resources in the reply below. #WorldModels #EmbodiedAI #Robotics #CausalAI

World models have a causality problem. Realistic videos are not enough. A world model should predict the future caused by an action, not just a plausible future. We find that many latent-action world models generate convincing videos while barely responding to the supplied action. The root cause lies in Latent Action Models (LAMs): reconstruction objectives entangle action-induced dynamics with action-irrelevant visual changes, producing visually confounded latent actions. Our solution is CD-LAM (Causally Debiased Latent Action Model). It cleanses latent action representations before pre-training begins, requiring zero changes to backbone architectures, latent dimensions, or action interfaces. Key Benchmark Results: - Action Controllability: Cuts action-following error by >30% while simultaneously enhancing visual fidelity. - Extreme Sample Efficiency: Achieves baseline performance in just 3,000 post-training steps instead of 50,000 - a >10x speedup. By removing visual confounding, CD-LAM bridges the gap between passive video prediction and actionable causal intelligence for robotics. The next bottleneck for world models isn’t realism. It’s causality. Paper, project page and additional resources in the reply below. #WorldModels #EmbodiedAI #Robotics #CausalAI

Biwei Huang

15,084 次观看 • 3 个月前

California startup announces breakthrough in general-purpose robotics with π0.5 AI — a vision-language-action model. Everything the robot sees, it sees for the first time. The robot has not been trained to recognize its surroundings. We are getting there! Only a few more years until we finally have household robots. A dream is coming true!

California startup announces breakthrough in general-purpose robotics with π0.5 AI — a vision-language-action model. Everything the robot sees, it sees for the first time. The robot has not been trained to recognize its surroundings. We are getting there! Only a few more years until we finally have household robots. A dream is coming true!

Chubby♨️

218,566 次观看 • 1 年前

🤖ROBOTS ARE GETTING SMARTER AT TOUCHING THE REAL WORLD Researchers from UC Berkeley, NVIDIA and Stanford introduced T-Rex, a framework that combines vision, language and tactile sensing. Instead of relying on cameras alone, robots can now respond to physical contact in real time. Robots are no longer just seeing objects. They’re learning how to FEEL them.

🤖ROBOTS ARE GETTING SMARTER AT TOUCHING THE REAL WORLD Researchers from UC Berkeley, NVIDIA and Stanford introduced T-Rex, a framework that combines vision, language and tactile sensing. Instead of relying on cameras alone, robots can now respond to physical contact in real time. Robots are no longer just seeing objects. They’re learning how to FEEL them.

Coin Bureau

28,275 次观看 • 4 个月前